A standard satellite navigation receiver determines position to within a few metres. That is fine for a car, but not for a mobile robot that has to stop at a specific component or drive along a narrow corridor. This is where RTK (Real-Time Kinematic) comes in - a method that achieves centimetre accuracy outdoors.
How RTK works
A stationary base station with precisely known coordinates stands near the work area. It sees the same satellites as the moving receiver (the rover), calculates the signal errors and sends corrections in real time. The rover applies them and determines its position to within centimetres. Corrections are usually transmitted over the internet via the NTRIP protocol or by radio.
- RTK Fixed - the best mode, centimetre accuracy.
- RTK Float - corrections are received, but accuracy is only at decimetre level.
- Without corrections - the usual metre-level accuracy.
Own base station or network?
Corrections can come from a national or commercial VRS network, or you can set up your own base station. Your own station close to the work area gives a more stable result and does not depend on the network. In our mobile platform project we use both: our own base station (based on a Septentrio mosaic-X5, 160-200 m from the rover) and a backup VRS network to which the system switches automatically if corrections drop out for more than a minute. We measured and corrected the offset between the two sources - they agree to within 4 mm.
Not just position, but heading too
A robot needs to know not only where it is, but also which way it is facing. Two antennas are used for this: the heading is calculated from the difference between their positions. On our platform the antennas are mounted 1.28 m apart, and the heading error is approx. 0.23° (1σ). If the antenna heading drops out briefly, an inertial sensor (gyroscope) holds it.
What the tests showed
- Static for 60 s: 600 out of 600 measurements in RTK Fixed mode, horizontal error approx. 9 mm (1σ) and 2 cm (95%).
- Driving along the route (10 Hz, Stanley controller): median deviation 5.2 cm, 95% of the time within 12.1 cm. This already includes the control accuracy of the platform itself.
- Near tall objects, some satellites are blocked and the signal is reflected. Raising the elevation mask to 12° (rejecting satellites low above the horizon) increased the RTK Fixed share next to the wagons from about 50% to 93-98%.
These are test results at a specific site, not accuracy proven in operation under any conditions.
Where it is useful
- Mobile robots and platforms working outdoors: inspection, measurement, transport.
- Precise stopping at an object, after which a robot with 3D vision takes over.
- Measuring large objects and sites where data has to be tied to a location.
It is important to understand that RTK brings the robot to the spot to within a few centimetres, while millimetre accuracy comes from the 3D scanner and robot path correction from the scan. We describe the complete system in our article on the autonomous inspection platform.
If you are planning a mobile robot or platform with precise navigation, get in touch.